Thanks for the suggestion.
I added
PARTITIONED BY (partition_date)
but not all of the data is loaded.
The parquet files are not stored in Hive-style partitioning. The folder format is
s3://bucket_name/send_date_summary/report/yyyy/mm/dd/*.parquet
where yyyy/mm/dd is the partition_date.
Do I need to manually add each partition? There are hundreds.
Is there an easier way to do this?
And will setting partition log automatically discover new partitions that are added to the folder?
SET spark.databricks.nonDelta.partitionLog.enabled = true;